The Q&A: Kay Wijekumar

With each issue, Trib+Edu brings you an interview with experts on issues related to public education. Here is this week's subject:

Kay Wijekumar is a professor of Teaching, Learning And Culture at Texas A&M University. Wijekumar developed software called the Intelligent Tutoring System that helps grade school students to better comprehend content area reading. She recently received a $3.5 million grant from U.S. Department of Education to continue her research. 

Trib+Edu: Could you explain what the Intelligent Tutoring System is and the role it plays in the classroom?

Kay Wijekumar: Our Intelligent Tutoring System is designed to improve content area reading comprehension — those texts that children will typically read in science and social studies and current events or sports. All of those are considered content area, and most children seem to have a great deal of difficulty in reading and understanding those content areas.

Actually, in fourth and fifth grade it’s been shown and it’s been noted that there’s something called a fourth grade slump, where kids who typically read stories from kindergarten to third grade come into fourth grade and suddenly experience a huge drop in their reading comprehension for school because they are unable to process and understand and remember the content area text. Content texts are different than typical stories. They don’t have plots and they don’t have characters, and so on. Instead they have very specific domain knowledge, so they might actually be missing the vocabulary and they might actually be missing the strategies by which they can process, read and understand the information that’s being presented.

So the Intelligent Tutoring System focuses on content area reading comprehension, and, specifically, we show children how to read and comprehend using text structure as a means to select and import information inside their brain. And we also show them how to summarize using text structure, how to infer using text structure and how to elaborate using text structure.

There are very specific strategies that are typically used in language art classes, but we actually show the children how to do those a lot more efficiently and effectively using text structure. And we teach children five text structures, which organize all of the content area texts in the world.

One is comparison; some people call it compare, contrast. The second is problem and solution. The third is cause and effect. Problem and solution and cause and effect are very tightly linked together because most problems have a cause or multiple causes and so on. Then the last two are sequence and description text structures. So we show children how to use five of the text structures and even master the text structures to read and comprehend information better.

Trib+Edu: How did you get the idea and how did you go about presenting this content learning system using technology?

Wijekumar: Early in the 1970s, actually, my co-principal investigating this project, Bonnie Meyer, had created this tech structure approach to reading comprehension and she would actually even teach adults to read and comprehend using this and their comprehension scores would go up. So in the '80s and '90s she started actually trying to build these into the language arts curriculum types of materials, and in fact the national reading panel mentioned text structure as one of the best approaches to improving reading comprehension, one of many approaches.

She had initially thought about extending the reach of the approach. The funding agency was kind of reluctant because they didn’t think it would be as impactful if we were just doing professional development for the teachers because there could be so many variations in teacher understanding of the text structure.

So I came into the picture and we talked about it, and I suggested to her that we build an Intelligent Tutoring System that will deliver this instruction and uniformly reach large numbers of students. And very carefully we could actually instruct the children, give them a lot of interactive practice in using the text structure, give them immediate feedback and do this one on one, which would actually improve the chances that the children are actually able to learn it. Whereas in the classroom setting the teacher might have 25 or 30 kids in the classroom — they couldn’t attend to each child and they would really have to understand how to use this text structure in their classroom. We thought that building an Intelligent Tutoring System gave us the best opportunity to reach as many students as possible, very consistently and in a high quality manner without all of the factors coming into play.

Trib+Edu: Your research emphasizes using this system in low-income schools. Why is this?

Wijekumar: We actually received funding from 2008 to 2013. We tested the Intelligent Tutoring System in fourth, fifth, seventh and eighth grade classrooms, approximately 130 classrooms at each grade level. We did this large, randomized control trial which is the gold standard for research and we found statistically significant, meaningful effects on both improvements in standardized tests as well as other measures like summarizing and so on. We did that in rural and suburban schools and we found these very positive effects and very important and meaningful results for most researchers and for practitioners.

We also, during that time, realized there are some very specific needs in high poverty schools because high poverty schools are known to have so many problems, including high teacher turnover. There is maybe inconsistent technology support. The children might actually have different types of needs. Prior knowledge might be different, and so on. So we wanted to also test the software and make sure that we adapt and create information for the children in high poverty schools particularly. So that’s why we sought funding and we were funded to actually continue the research and to do the new study at high poverty schools.

Now in the meantime, in the last three years, we actually created an adaptation of our system for Spanish-speaking English language learners. In the Intelligent Tutoring System, typically, the kids will actually hear a talking head, his name is IT, and IT will say, "Today I’m going to teach you how to read about two presidents. These two are very different. One is Abraham Lincoln and the other is George Washington. I’m going to show you how to remember more from what you read.”

So this original Intelligent Tutoring System is built completely in English with lots of support for vocabulary and things like that. But then we sought funding and we built an adaptation for Spanish-speaking students who would actually be able to see the information and read it in English. But, if they are unclear or unsure of a word, they could hover over the word and they could see it translated into Spanish.

So we were giving them a little bit of a system in their primary language to help them make an even clearer understanding in English. Even if it’s a slightly more advanced version of the Spanish tutor, that also will present every page in Spanish followed by English, because they would see it in their primary language and they could then extend it to English. So little by little they could actually learn how to read in English by first seeing it in Spanish.

So we have two versions that we created, and we just finished testing. So when we go into high poverty schools in Texas and other states that are in the southern part of the United States, we expect that some children who are still transitioning will be able to use that software in addition to our English version, for the majority of the kids.

Trib+Edu: How did you land on focusing this system on elementary-aged students?

Wijekumar: We focus it on elementary, middle and high school-aged students, but there’s lot of research that shows that you should try to help children as early as possible, as young as possible so that they can have all of the strengths of using these strategies as they move forward into middle and high school. So our goal was to try and address these very specific needs as early as possible. That’s why we started at fourth and fifth grade, so that we could prepare the kids to be successful in middle and high school.

Trib+Edu: Can you explain the benefits of using the different text structures emphasized in the system: comparison, problem and solution, cause and effect, sequence, and description?

Wijekumar: In the typical language arts classroom, when you ask a child to summarize, they will actually have a little bit of difficulty. To give them a passage that compares two presidents, for example, the children will say, “Oh it was one president” or they will say, “Oh it was about Lincoln.”  They will tend to engage in this knowledge-telling approach, like think of things and paraphrase and they don’t actually summarize in the text.

Our approach is unique in that we tell the child when they see signaling words like "in contrast", "similarly", "on the other hand" and they see signaling words or linking words like that it should tell them that something or someone is being compared. So given that, there is a comparison going on here, we ask the child then to say, OK, let me summarize. The specific pattern by which we summarize, this pattern is really unique to what we show the children.

So in the comparison text structure, we show the children a pattern “blank and blank were compared on blank and blank and blank.” So we ask, who was being compared? On what basis were they being compared? This is a lot more transparent and really easy for the child to follow, instead of telling the child “well, read it again, what is important?” In our approach, we actually tell the children how to summarize, given the text structure.